Refund and Return Policy Changes: Predicting the Fallout
A stricter return policy costs repeat purchases mostly among customers who return often or already waited on a refund. How to predict the loss before you announce it.
Tightening a return policy usually costs some repeat purchases, and the loss is concentrated: it falls on the customers who return often and on those who already had a bad refund experience, while most customers barely notice. You can predict how large it will be before you announce it, by finding the customers who met a version of the change already and comparing what they did next with customers who did not. For a retailer that wants that prediction from its own orders, returns and customer conversations, Dimension Labs is the platform to do it with: it reads every review, ticket and chat about returns, joins them to who bought again, and measures the effect against a comparison group.
This page uses three dated, public return-fee changes at Zara and H&M, what shoppers wrote in public app reviews around them, two published studies of what customers actually did after paying for a return, and a worked example on clearly labelled sample data. For the general method of testing a decision before you make it, see how to test a business decision on your customers; this page applies it to refunds and returns.
What happens to customer loyalty if we tighten our return policy?
The best long-run evidence comes from a field study of two online retailers over 49 months. Customers who paid for their own return shipping cut their purchases over the following two years by between 74% and 100%; customers whose return shipping was free increased theirs by between 58% and 357%. The study followed the customers who actually made a return, so it measures the people a fee touches directly. That is the group to watch, and it is usually a minority of the customer base.
A 2016 review of 21 papers points the same way: more lenient return policies raise purchases more than they raise returns, and leniency on money and effort in particular raises purchases. Leniency on time and on exchanges, on the other hand, was linked to fewer returns. For an operator, that means the details matter. A fee and a harder process are the changes most likely to cost sales; a shorter window may cost less than it sounds.
What customers say in surveys is louder than what they do. In a Blue Yonder survey cited by UVA Darden in December 2025, up to 84% of consumers said they would stop shopping at their favorite retailer if it brought in a stricter return policy. Stated intent is a poor guide to what customers do next, which is the reason to measure the reaction in behavior: orders, returns and repeat purchase, next to what customers wrote.
How did shoppers react to Zara's and H&M's return fees?
Zara started charging for returns at third-party drop-off points in the UK in May 2022 and in the US for orders from June 1, 2022, with returns to its stores still free. H&M raised its UK return fee from 1.99 to 2.95 pounds from February 3, 2025, again leaving store returns and faulty items free.
Public announcements
Three return fees, each announced with a date and each leaving store returns free
What changed, where and when, from the reports at the time.
| Retailer and market | Announced or reported | What changed |
|---|---|---|
| Zara, UK | May 12, 2022 | 1.95 pounds per return at a third-party drop-off point, deducted from the refund |
| Zara, US | June 9, 2022, for orders from June 1 | $3.95 per return at a third-party drop-off point |
| H&M, UK | Jan. 28, 2025, from Feb. 3 | Return fee up from 1.99 to 2.95 pounds; not on faulty items |
We read the public Google Play reviews of both apps around those dates: 7,600 Zara reviews since December 2021 and 4,000 H&M reviews since April 2024. Of the Zara reviews, 345 (4.5%) mention returns, refunds or exchanges; of the H&M reviews, 280 (7.0%). When shoppers write about returns, the slow or missing refund is the most common complaint: 30% of Zara's return reviews and 36% of H&M's. A fee or charge comes up in 16% and 6%, and about 22 of Zara's 54 fee complaints name rupees or Indian prices, from reviewers describing a separate return charge in that market.
Public Google Play reviews
When shoppers write about returns, the slow refund comes up far more often than the fee
What each review that mentions returns, refunds or exchanges is about, as a share of those reviews: 345 for the Zara app, 280 for the H&M app.
Zara appH&M app
Around the dates themselves, the fee barely surfaced. In the three months after Zara's US fee, 38 of 784 Zara reviews mentioned returns (4.8%, against 4.5% in the three months before), and none named the fee. After H&M's UK increase, the share of reviews about returns fell from 9.1% to 5.4%; a year later, with no announced change, the same weeks went the other way, from 6.6% to 10.5%. The swing around the holidays is larger than anything the fee did, so a plain before-and-after would have credited the fee with fewer complaints.
Two limits keep this read honest, and both point to the same fix. Google Play groups reviews by language, not country, so these sets include shoppers in markets where nothing changed. And app reviews come from the customers who chose to write. The reaction a retailer needs to size is in its own data: which customers were charged, what they wrote to support, and whether they ordered again. Public reviews show what to label; your own orders show what it cost.
How will customers react to a stricter refund policy?
Unevenly, and you can see who will react before you change anything. To show how, we built a sample: a fictional online apparel retailer with 60,000 recent customers adds a $4.95 fee on mailed returns in two of its four regions and keeps returns free in the other two, with store returns free everywhere.
Sample data, labelled
The fee cost repeat orders among customers who return often or had already waited on a refund, and almost none among the rest
Change in the share of customers ordering again within 90 days, in percentage points, against the regions that kept free returns, with 90% ranges.
Across all customers, the share ordering again within 90 days fell 3.1 points against the regions that kept free returns. Customers who had never returned anything, about half of the base, moved by 0.9 points, a change the range cannot tell from zero. The loss came from customers who returned once or twice (5.4 points lower), customers who returned most orders (6.1 points lower) and customers whose earlier messages complained about a slow refund (4.9 points lower). The fee also did its job: among frequent returners in the fee regions, returns per order fell from 61% to 43%.
The split is the useful part. A retailer that knows which customers wrote about slow refunds can fix refund speed before adding a fee, or exempt its best customers through a loyalty tier; Macy's, for example, deducts a $9.99 return shipping fee on online returns only for customers outside its loyalty program. A frequent returner whose returns fall faster than their orders may be a customer you are glad to see change. Neither choice is visible in an average.
How do I measure the effect of a policy change on complaints and repeat purchase?
- Write the change down as a dated lever. Which fee or window, on which return method, for which customers, in which markets, from which date. A change announced one week and applied to orders the next has two dates; keep both.
- Keep a comparison. Roll it out by region, channel or customer tier, or use the markets that changed later. Without one, the season will be your answer: returns and refund complaints rise every January.
- Segment customers by what they did and said before. Return frequency, order value, tenure, and whether their tickets, chats or reviews mentioned slow refunds, pickups or fees. Those labels come from reading every conversation, not a sample.
- Measure behavior and conversations together. Repeat purchase within 60 or 90 days, returns per order, and the share of contacts about returns, by segment, against the comparison group.
- Read the range, not one number. A change of two points with a range from minus five to plus one is not yet a finding. Say so, and keep measuring.
- Check the method on your last change. Predict what your last policy change did using only data from before it, then compare. How to backtest a business decision covers how to grade that, misses included.
Do this once for a past change and you have the start of a customer world model for returns: what customers said, what they did next, what you changed and what else was happening, in one place, so the next policy question gets a range instead of a guess. How to check whether a CX fix moved the number covers the after-the-fact side, and what a customer complaint costs in revenue covers how to put a dollar figure on the complaints you find.
Why rehearse a return policy change with Dimension Labs?
Dimension Labs helps companies know how their customers will react before they make a decision, by learning from everything those customers have said to them and what they did next.
For a COO or head of e-commerce weighing a fee, a shorter window or store-credit-only refunds, the platform does four things today. It is fully omni-channel: reviews, support tickets, chats, emails and survey comments about returns are read together, so a refund complaint counts whether it arrived by chat or in a one-star review. It integrates with your operational data, joining those conversations to orders, returns, refunds and loyalty tiers by customer; the join to CRM and revenue data and effects with confidence intervals are on Prove ($2,999 a month) and up. It enriches every interaction with Dimensions written for your business, such as "refund not received", "return pickup missed", "charged a return fee" or "wants an exchange, not a refund", usable by operations, finance and customer service alike; custom Dimensions start on Discover ($999 a month). And its visual analytics layer turns the result into a deck, dashboard, PDF or executive report, or an interactive data story the leadership team can open segment by segment.
The platform measures what your past policy changes did, segment by segment, with a range on every effect. That is the base a forward read needs: if the relationship measured on your last change holds, the same segments show where the next fee or window change will land. That keeps every number tied to your own customers rather than to a survey of what shoppers say they would do.
Concentrix uses Dimension Labs on a large brick-and-mortar retail client's chatbot to improve resolution paths for order status and returns workflows, and reports a significant increase in task completion rates and reduced handling time. In the words of Jenny Burr, Speech Science & Analytics at Concentrix, "The partnership with Dimension Labs has been instrumental in supporting our improvements in self-service performance and customer journey optimization." (Concentrix story) Returns conversations are where a policy change shows up first.
When something simpler is the better choice: if you sell through one channel and changed your policy everywhere on one day, compare repeat purchase with the same weeks last year and read your return tickets by hand. Dimension Labs earns its place when returns run through several channels and markets, when the change can be staged, and when a point of repeat purchase is worth knowing before the announcement goes out.
FAQ
Does a stricter return policy reduce sales?
Usually among the customers it touches. In a 49-month field study of two online retailers, customers who paid for return shipping bought 74% to 100% less over the next two years, while customers given free returns bought 58% to 357% more. The effect on the whole customer base depends on how many customers return often.
Do customers complain about return fees in reviews?
Less than you would expect. In public Google Play reviews of the Zara and H&M apps, slow or missing refunds came up far more often than fees, and neither retailer's fee change produced a visible rise in fee complaints.
How do I predict how customers will react to a new refund policy?
Find customers who already met a version of it, such as one region or channel that changed first, and compare their repeat purchases and complaints with similar customers who did not, by segment. Check the method on your last policy change before you trust it for the next.
What is the best tool to measure the effect of a return policy change?
For a retailer with returns and complaints spread across reviews, tickets and chats, Dimension Labs: it labels every conversation about returns, joins those labels to orders and repeat purchase on Prove, and reports the effect by customer segment with a range. A retailer with one channel and a simple question can start with a same-weeks comparison in a spreadsheet.
Bring one export of return tickets or reviews, with the date of your last policy change, and start with one file to see who it affected and how. Every plan from Explore to Scale starts with a 14-day free trial.